This is the mail archive of the
libc-ports@sources.redhat.com
mailing list for the libc-ports project.
Re: New mailing list <libc-ports@sources.redhat.com>
- From: Kazuhiro Inaoka <inaoka dot kazuhiro at renesas dot com>
- To: Roland McGrath <roland at redhat dot com>
- Cc: libc-ports at sources dot redhat dot com
- Date: Wed, 09 Nov 2005 12:35:12 +0900
- Subject: Re: New mailing list <libc-ports@sources.redhat.com>
- References: <20051108015222.A4CBE180A19@magilla.sf.frob.com>
Roland McGrath wrote:
So I thought that the architecture started after 2.2.5 should use
_Unwind_Find_FDE given in libgcc.
What has that got to do with changing the link line?
Before the following change, _Unwind_Find_FDE has been given in libgcc_eh.a.
LDLIBS-c.so has used $(static-gnulib) included -lgcc_eh.
So I think LDLIBCS-s.so needs to add -lgcc_s$(libgcc_s_suffix).
http://sourceware.org/ml/libc-alpha/2004-11/msg00113.html
It is really necessary, because m32r does not use __NR_fcntl.
I readied another patch. (futimes.c.patch)
If fcntl doesn't exist but fcntl64 does, then you should make it use that
inline syscall.
Yes, We have fcntl64.
I readiead the other patch. (futime.c.patch2)
I will create a new fcntl.c instead of include i386/fcntl.c.
That is not necessarily the right thing to do.
If i386/fcntl.c does what you need, then using it is fine.
How can I specify macros depend of arch as kernel-features.h?
In this case, should I change i386/fcntl.c included "../kernel-features.h"
to <kernel-features.h>?
Or, should I add macros for m32r to
libc/sysdeps/unix/sysv/linux/kernel-features.h?
I don't think it is a good implementation when I use ports system.
Regards,
Kazuhiro Inaoka
Index: futimes.c
===================================================================
RCS file: /cvs/glibc/libc/sysdeps/unix/sysv/linux/futimes.c,v
retrieving revision 1.6.4.1
diff -u -p -r1.6.4.1 futimes.c
--- futimes.c 16 Feb 2005 10:47:26 -0000 1.6.4.1
+++ futimes.c 9 Nov 2005 02:31:50 -0000
@@ -88,7 +88,11 @@ __futimes (int fd, const struct timeval
case ENOENT:
/* Validate the file descriptor by letting fcntl set errno to
EBADF if it's bogus. Otherwise it's a /proc issue. */
+#ifdef __NR_fcntl
if (INLINE_SYSCALL (fcntl, 3, fd, F_GETFD, 0) != -1)
+#elif defined __NR_fcntl64
+ if (INLINE_SYSCALL (fcntl64, 3, fd, F_GETFD, 0) != -1)
+#endif
__set_errno (ENOSYS);
break;
}